Persistence of Apache Trout Following Wildfires in the White Mountains of Arizona

PERSISTENCE OF APACHE TROUT FOLLOWING WILDFIRES IN THE WHITE MOUNTAINS OF ARIZONA Jonathan Long Wildfires are a natural disturbance that can Mountain suggest that this landscape has rejuvenate and rebuild trout habitat by stim- experienced stand-replacing wildfires, ulating flood scour and deposition (Gress- because extensive stands of aspen (Populus well 1999; Bisson et al. 2003). However, tremuloides) often reflect past wildfires in several case studies from Arizona and New mixed-conifer forests (Swetnam et al. 2001; Mexico have shown severe reductions in fish Margolis 2003). However, a recent study of populations shortly after wildfires (Neary et stand-replacing high-elevation fires (Mar- al. 2005). The consequences of wildfire are a golis 2003) did not include the White Moun- particular concern in the mountainous wat- tains, in part because other areas offered ersheds that harbor Apache trout (Oncorhyn- clearer historical evidence of such fires (Ellis chus gilae ssp. apache) and Gila trout (Onco- Margolis, personal communication). rhynchus gilae ssp. gilae). Populations of these Recent climatic conditions and historical federally protected species are scattered management practices have increased the among small stream reaches that are isolated likelihood of large and severe wildfires in by the presence of non-native trouts, inter- the White Mountains. Widespread insect mittent flows, thermal regimes, natural infestations, abetted by warm winters and barriers, and barriers to exclude non-native dry summers, have killed huge swaths of trout (Brown et al. 2001; Rieman et al. 2003; spruce in bands around Mount Baldy USFWS 2006). Wildfires have extirpated (Lynch 2004). The combination of increased seven populations of Gila trout since 1989, forest fuels and warming temperatures over impeding recovery efforts (Propst et al. 1992; the past century has increased the likelihood USFWS 2006). Emergency evacuation of of wildfires across vegetation zones in Ari- populations threatened by fire has become zona mountains (Cocke et al. 2005). Re- an important element of Gila trout recovery searchers have asserted that past timber (Anderson 1992; USFWS 2006), and has been harvest and road construction activities have recommended as a “fundamental manage- rendered habitats more vulnerable to post- ment approach” for conserving isolated pop- fire impacts; this combination of threats has ulations of rare native fishes (Rinne 2004). motivated many land managers and com- However, because evacuation is costly and munity members to express concern that stressful to fish, managers need to know large wildfires in the White Mountains when it is necessary to prevent population could jeopardize valued resources such as losses. Apache trout (Abrams 2005). Debates over The extent of the threat to Apache trout how to manage forests in the face of such from wildfire has not been clear, since large risks need to be guided by analyses of wild- wildfires have not struck its ancestral home fire effects within local landscape contexts on Mount Baldy in the White Mountains of (Rieman et al. 2003). Arizona in several decades (Dieterich 1983; Two streams (Grant Creek and KP Creek) Gomez and Tiller 1990). Place names such as whose high-elevation watersheds have Aspen Butte, Aspen Ridge, and Burnt recently experienced wildfires (the Steeple 220 Jonathan Long fire of July 2003 and KP fire of May 2004) are (Schaffner and Reed 2005). Cannon et al. especially important because they represent (2003) found that slopes greater than 30 per- ancestral habitat for native trout (Figure 1). cent and soils with increased permeability Native trout had been collected from KP and organic matter content were associated Creek in 1904, although the taxonomic with increased probability of fire-induced identity is uncertain because the original debris flows. Debris flows are also strongly specimens have been lost (Miller 1950). KP related to the occurrence of post-fire storm Creek was later stocked with a strain of events (Cannon 2001), but it is difficult to Apache trout from an unconnected stream predict such events. The availability of light- as part of recovery efforts (Dowling and ly burned refugia and absence of barriers are Childs 1992). Researchers have concluded two factors that help fish avoid impacts that the present-day trout populations in from fires (Burton 2001; Bisson et al. 2003). both KP Creek and Grant Creek are hybrid- Previous research in the White Moun- ized with rainbow trout (Figure 2; Rinne tains indicates that landscape attributes such 1985; Dowling and Childs 1992; USFWS as topography, geology, and vegetation type 2003). The current recovery plan is to stock need to be considered when evaluating fish these streams with the Spruce Creek lineage habitat (Long and Medina 2005). Within a of Gila trout, which appears to be an inter- region, mean basin elevation and slope help mediate form between Gila and Apache to explain variation in post-fire flooding trout (Riddle et al. 1998; USFWS 2003). The (Schaffner and Reid 2005). Therefore, com- impact of wildfire on the two creeks was paring landscape attributes of the study thus not a major concern from a conserva- streams to those designated for Apache tion standpoint, but their response to wild- trout recovery would help in evaluating fires could yield insights into how wildfires whether their post-fire responses would be might affect other streams that have been representative. Extrapolating the responses designated for recovery of Apache trout. of two streams to a much larger and more The severity of a burn within a stream’s diverse region is speculative, but it could watershed is a primary factor in predicting help managers who must act on limited the impact of fires on aquatic systems. information concerning the risk of wildfires Burton (2001) found that 60–88 percent of to Apache trout. the watersheds of streams that experienced high levels of post-fire debris flooding had METHODS burned at high severity; only 14–37 percent This study examined the upper reaches of of the watersheds of streams that did not KP and Grant Creeks, perennial tributaries experience post-fire debris flooding were of the Blue River (Figure 1). A five-person burned at high severity. Cannon (2001) has team from the Rocky Mountain Research reported that the percent of watershed area Station (RMRS) sampled fish populations burned at moderate to high severity is an and habitat conditions at seven 50 m long important predictor of debris flows, and sampling sites in KP and Grant Creeks in Schaffner and Reed (2005) successfully pre- June of 2004 (Figure 3), after the KP fire was dicted increases in peak flows by calculating contained. Fish populations were resampled area burned at moderate to high severity. at six of the sites one year later. The team Managers can easily evaluate burn severity attempted to relocate sites that had been in the aftermath of a wildfire, making it an sampled for fish in September of 1995 by the important and accessible tool for predicting Arizona Game and Fish Department fish persistence in different settings. (AGFD) as part of their General Aquatic A variety of factors besides burn charac- Wildlife Surveys (GAWS) program. The teristics affect the responses of stream eco- team successfully located three of those systems to wildfires, including human monitoring sites (labeled 22, 20, and 19 from influences and natural landscape attributes upstream to downstream) in the upper Jonathan Long 221 Figure 1. Locations of Apache trout streams and the study streams in the White Mountains of east- central Arizona. reaches of Grant Creek, although AGFD had with nets to prevent fish from escaping sampled fish only at the middle site. In KP during sampling and each was sampled Creek, the original site markers could not be three times using the depletion method located, so the team approximated the three (Zippin 1958). Lengths of all fishes were uppermost sites (24, 23, and 22 from up- recorded and used to calculate mean lengths stream to downstream) based on the with 95 percent confidence intervals. The descriptions of channel gradient and intra- maximum likelihood population estimate station distances recorded on the 1995 and 95 percent confidence interval were survey forms. Fish populations were sam- calculated at each sampling site using the pled using backpack-mounted electro- Microfish 3.0 computer program (Van shocking gear. Each reach was blocked off Deventer and Platts 1989). 222 Jonathan Long Figure 2. Hybridized Apache trout collected in KP Creek, June 2004. Land Resource Information System served Landscape Attributes of Study Sites to characterize site geology (Figure 4). The Watersheds of the study sites were deline- two major geologic types are felsic volcanic ated using the hydrologic modeling tools in formations along the upper slopes of Mount ArcMap version 9 (ESRI, Redlands, CA). Baldy and mafic volcanic formations along Burn severity information (based on differ- the lower slopes and adjacent plateaus. enced normalized burn ratio data calculated Slopes were calculated using ArcMap to from satellite images) provided by the process 10 m digital elevation models of the Apache-Sitgreaves National Forest was region (Figure 5). Data for the vegetation incorporated into the geographic database, layer were provided by the U.S. Geological and the data were used to calculate the per- Survey’s GAP analysis program. In the three cent of each watershed burned at moderate major mixed-conifer forest types that pre- to high severity. dominate within the range of Apache trout Mapping landscape attributes of KP and (Figure 6), the dominant tree species are Grant Creek served to compare them to the Engelmann spruce (Picea engelmannii), Doug- 36 streams that the Apache Trout Recovery las fir (Pseudotsuga menziesii), and ponderosa Team has identified as priorities for conser- pine (Pinus ponderosa). vation of pure strains of Apache trout within Maps and elevations in the draft recovery its ancestral range (Ruiz and Novy 2000).

Load more